Efficient gene transfer into human primary blood lymphocytes by surface-engineered lentiviral vectors that display a T cell–activating polypeptide

Efficient gene transfer into T lymphocytes may allow the treatment of several genetic dysfunctions of the hematopoietic system, such as severe combined immunodeficiency,1,2 and the development of novel therapeutic strategies for diseases such as cancers and acquired immunodeficiency syndrome.3 To reach this goal, it is essential to preserve the functional properties of the transduced cells, that is, their capacity to appropriately react on stimulation of the immune system. Therefore, only minimal ex vivo manipulation of the cells should be performed during gene transfer because growth factor/cytokine combinations, used to force cell proliferation, often lead to skewing of cell populations and may alter their ability to respond to novel antigens.4,5 Ideally, gene delivery should be best achieved in vivo to minimize contacts of target cells with nonphysiologic cell culture reagents that may induce differentiation or proliferation.

Vectors derived from retroviruses are probably among the most suitable tools to achieve a long-term gene transfer because they allow stable integration of a transgene and its propagation in daughter cells. To date, vectors derived from oncoretroviruses such as murine leukemia viruses (MLVs) have been widely used for gene transfer into human T cells,6 essentially because of the simplicity of their manipulation. Perhaps one of the most important drawbacks associated with the use of such vectors is their inability to transduce nonproliferating target cells. Indeed, following internalization of the vector into the target cell cytoplasm and reverse transcription, transport of the preintegration complex to the nucleus requires the breakdown of the nuclear membranes during mitosis.7,8 This provides a formidable barrier to the use of MLV-based vectors in the many gene therapy protocols for which target cells are quiescent or for which induction of cell proliferation is to be avoided. Thus, the recent emergence of lentiviral vectors may provide a valuable alternative to overcome this problem owing to the lentivirus mechanism that allows mitosis-independent nuclear import of the preintegration complex and infection of nonproliferating cells.9-11 

Several studies have now established the capacity of these vectors derived from human immunodeficiency virus 1 (HIV-1) to transduce various types of nonproliferating cells both in vitro and in vivo.12 However, some cell types that are important gene therapy targets are refractory to gene transfer with lentiviral vectors, despite the most recent improvements brought into their structures.13-16 This includes, in particular, early progenitor hematopoietic stem cells in G₀,17monocytes,18,19 and resting T lymphocytes.14That the parental virus, HIV-1, can enter into resting T lymphocytes but does not replicate,20-24 has been attributed to multiple post-entry blocks. This includes, in particular, (1) defects in initiation and completion of the reverse-transcription process,20,23-25 (2) lack of adenosine triphosphate-dependent nuclear import,9,26 and (3) lack of integration of the proviral genome.27 Low levels of nucleotides in the resting cells do not entirely explain the restricted HIV-1 replication because artificially raising intracellular nucleotide pools increased reverse-transcription products but not the level of productive infection.28 However, it was recently reported that inducing the resting T cells to enter into the G_1bphase of the cell cycle by stimulation through the T-cell receptor (TCR) and CD28 costimulation receptor, using anti-CD3 plus anti-CD28 antibodies, was sufficient to render the cells susceptible to HIV-1 infection and replication.25 Moreover, exposing T cells to cytokines that do not trigger cell division could render them permissive to transduction with HIV-1 vectors.29 These findings suggest that partial activation of resting T cells is sufficient for gene transfer by HIV-1–derived vectors and that DNA synthesis or mitosis of these cells is not necessary.

Here we sought to refine the structure of lentiviral vector particles to overcome their inability to transduce nonactivated T cells. Because extensive activation of resting T cells with combinations of stimulating factors should be avoided ex vivo and may not be possible to achieve in vivo, we have designed novel lentiviral vector particles that transiently provide a minimal stimulus to the target cells during gene delivery. As a proof of concept, we establish here that lentiviral vectors whose virion surface was engineered to display a TCR-activating polypeptide could efficiently transduce primary resting T lymphocytes without addition of soluble T-cell activation stimuli in the cell culture medium. HIV-1–derived vectors were copseudotyped with an MLV-derived chimeric envelope glycoprotein fused to an anti-CD3 single-chain antibody variable fragment (scFv) and the vesicular stomatitis virus G (VSV-G) glycoprotein. This anti-CD3 scFv was derived from the OKT3 monoclonal antibody, which activates the TCRs. Importantly, the surface-modified lentiviral vectors specifically recognized CD3 and triggered activation of freshly isolated peripheral blood T cells. This minimal stimulation was sufficient to allow gene transfer in up to 48% of the lymphocytes, that is, 100-fold more than the performance of unmodified lentiviral vectors in nonactivated T cells.

The 293T human embryo kidney cell line was grown in Dulbecco modified Eagle medium (DMEM; Life Technologies, Cergy Pontoise, France) supplemented with 10% fetal calf serum (FCS). The Jurkat human T leukemia cell line was grown in RPMI 1640 (Life Technologies) supplemented with 10% FCS. Human peripheral blood mononuclear cells (PBMCs) were separated from fresh blood of healthy donors using a Ficoll-Hypaque/Percoll gradient (Amersham Pharmacia Biotech, Orsay, France). Peripheral blood lymphocytes (PBLs) were enriched from the PBMC fraction by an overnight plastic adherence at 37°C to remove adherent monocytes. The nonadherent cells were resuspended in RPMI 1640 medium supplemented with 10% FCS. PBLs were checked for CD3 marker expression (75%-85% were CD3⁺) and the percentage of cells in the different phases of the cell cycle was determined by flow cytometry (≥ 99% were in G₀/G₁).

The pCMVΔ8.230 and pCMVΔ8.9131HIV-1 packaging plasmids have been described elsewhere. The HPPT-EF1α-GFP vector contains the EF1α internal promoter driving the GFP (green fluorescent protein) gene and the cPPT/CTS (central poly-purine track and central termination sequence) cis-acting signal, which stimulates nuclear import.15,16 The pCMVΔ8.2 packaging-deficient construct contains all HIV-1 genes except env, and pCMVΔ8.91 lacks the HIV-1 accessory genes vif, vpr,vpu, and nef. The CMV+intron plasmid, containing the MLV packaging sequences (S. Chapel-Fernandes and F-L. Cosset, unpublished material, 1996), and the MFG-GFP plasmid, encoding an MLV-based vector containing the GFP marker gene, were used to generate MLV-derived vectors. phCMV-G32encodes the VSV-G glycoprotein under control of the human cytomegalovirus (hCMV) promoter and rabbit β-globin intron II and polyadenylation sequences.

The scFv gene encoding a polypeptide that binds the human CD3 complex of TCRs was derived from the OKT3 monoclonal antibody33 and was described elsewhere.34 The OKT3 scFv complementary DNA (cDNA) was fused in the 4070A (amphotropic) MLV env gene,35 at a position corresponding to the first amino acid of the SU (surface) subunit of the amphotropic MLV envelope glycoprotein, as described previously.34 To reduce steric hindrance between the scFv polypeptide and the amphotropic envelope glycoprotein, a peptide linker, 7 amino acids long, was inserted between these 2 domains.34,36 The resulting chimeric glycoprotein was named OKT3SU (Figure1). The SUx mutation,37which inhibits furin-mediated cleavage of the MLV glycoprotein in the producer cells, was inserted in the OKT3SU construct, thus resulting in a second chimera, named OKT3SUx. This was achieved by replacing the Lys-Tyr-Lys-Arg furin cleavage site by the IPe-Glu-Gly-Arg peptide.37 Both the OKT3SU and OKT3SUx chimeras were expressed using the phCMV-G expression vector backbone.32 

Pseudotyped HIV-1–derived vectors were generated as previously described30 by transient transfection of 293T cells. Then 8.6 μg of the HPPT-EF1α-GFP vector construct, 8.6 μg of either pCMVΔ8.2 or pCMVΔ8.91 packaging constructs, and 2.7 μg of theenv construct, for example, the VSV-G–expressing construct (phCMV-G) were used to cotransfect 293T cells. When HIV-1–derived vectors were copseudotyped with VSV-G and OKT3SU or OKT3SUx chimeric glycoproteins, an equimolar quantity of envelope-expressing plasmids was used. A similar protocol was used to produce MLV-derived vectors with the CMV+intron and the MFG-GFP expressing constructs. Plasmid DNAs were transfected into 2.6 × 10⁶ 293T cells seeded the day before in 10-cm diameter plates using the calcium-phosphate transfection system according to manufacturer's recommendations (Life Technologies). The medium (12 mL/plate) was replaced 16 hours after transfection, and supernatant was harvested 24 hours later, low-speed centrifuged (2000 rpm for 5 minutes at 4°C), filtered through 0.45-μm pore-sized membranes, and directly used in the different assays.

To determine transduction efficiency and infectious titers of HIV-1–derived vectors, 293T target cells were seeded at a density of 2 × 10⁵/well in 6-well plates 1 day before transduction. Serial dilutions of vector preparations were added to the 293T cells. The percentage of GFP⁺ cells was determined by FACS analysis, following homogenization of transduced cells in trypsin and resuspension in phosphate-buffered saline (PBS). The titer was derived from the percentage of GFP⁺ cells after transduction of 3 × 10⁵ target cells with 1 mL viral supernatant. The infectious titers are expressed as 293T transducing units (TU/mL).

To infect human primary lymphocytes, 1 mL viral supernatant containing 5 × 10⁵ to 5 × 10⁶ TU, as determined on 293T target cells, was added to 1 to 3 × 10⁵ lymphocytes suspended in 0.5 mL RPMI supplemented with 10% FCS in 24-well plates. Multiplicities of infection (MOIs) were determined on proliferating 293T cells and are indicated in all lymphocyte transduction experiments. Human primary lymphocytes were infected in the same conditions as 293T control cells and infection efficiency was determined by flow cytometry at day 6. For T-cell stimulation and control infection assays performed with vectors pseudotyped with VSV-G only, anti-CD3 (HIT3a, Pharmingen) and anti-CD28 (CD28.2, Pharmingen) antibodies were used at a final concentration of 1 μg/mL/point. Activation with soluble anti-CD3/anti-CD28 antibodies was preferred to activation with immobilized (plastic-coated) antibodies, to provide a true comparison with the OKT3SU-displaying virions. This resulted in approximately 30% lower transduction efficiencies than obtained in transduction protocols that used immobilized OKT3 antibodies (data not shown).4,14,25,28 No differences in transduction efficiency were found when comparing OKT3 versus HIT3a anti-CD3 monoclonal antibodies for T-cell activation and infection (data not shown).

For the short exposure of PBLs to virus supernatants, the cells were washed with RPMI medium 24 hours after infection, resuspended in RPMI containing 10% FCS supplemented with or without recombinant interleukin (IL) 2 (rIL-2; 1 ng/mL; R & D Systems, Abingdon, United Kingdom) and maintained up to day 6 after infection for FACS analysis.

Anti-SU (Viromed Biosafety Labs, Camden, NJ) was a goat antiserum raised against the Rausher leukemia virus gp70, used diluted to 1:2000 for Western blots. Anti-MLV-CA (Viromed Biosafety Labs) was a goat antiserum raised against the Rausher leukemia virus p30 capsid protein (CA), used diluted to 1:10 000 for Western blots. Anti-VSV-G P5D4 monoclonal antibody (1 μg/μL; Sigma, St Louis, MO) was diluted to 1:1000 for Western blots. Anti-HIV-1-CA (p24 capsid) monoclonal antibody (8 μg/μL) was diluted to 1:8000 in immunoblots. For Western blot analysis, lysates of vector producer cells and virion samples were prepared as previously described.38 

Anti-CD3 was used at a 1:125 dilution for cell surface staining. Staining for CD25, CD69, and CD71 T-cell activation markers was performed with phycoerythrin-conjugated anti-CD25, anti-CD69, and anti-CD71 antibodies (BD-Pharmingen, Pont de Claix, France) at a 1:25 dilution. Target cells (5 × 10⁵ cells/point) were incubated with the appropriate antibodies for 45 minutes at 4°C. Fluorescence of living cells was analyzed with a fluorescent-activated cell sorter (FACSCalibur, Becton Dickinson, Pont de Claix, France).

Supernatant of the 83A25 hybridoma,39 secreting a rat monoclonal antibody against MLV SU, was used undiluted for binding assays. Binding assays on Jurkat T cells (5-8 × 10⁵cells/point) were performed as previously described.37 For competition binding assays, Jurkat cells were preincubated for 45 minutes at 4°C with an anti-CD3 antibody (HIT3a, 1 μg/μL, Pharmingen) diluted to 1:100 to saturate TCRs, then exposed to virus. Cells were then processed as previously described38 except that 10 ng/mL mouse IgGs (Sigma) was added with the secondary antibody to inhibit nonspecific antibody recognition.

Freshly isolated human PBLs (1-3 × 10⁵ cells) were maintained in RPMI supplemented with 10% FCS. PBLs were stimulated for different time periods after cultivation with 1 mL filtered viral supernatants or with stimulating agents (soluble anti-CD3 or soluble anti-CD28 antibodies). Cells were then centrifuged and fixed in 70% ethanol-30% PBS. The cells were stored at 4°C until FACS analysis. Just before FACS analysis, the fixed cells were pelleted and resuspended in 1 mL PBS containing 50 μg RNase and 10 μg/mL propidium iodide. The cells were incubated for 30 minutes at 37°C and the cell cycle was then analyzed with a fluorescence-activated cell sorter (FACSCalibur, Becton Dickinson).

Previous data from our laboratory have shown that the MLV envelope glycoprotein can be engineered to display various types of polypeptides at its amino terminus.40 Because the wild-type MLV glycoprotein is efficiently incorporated on HIV-1–derived lentivirus vectors,30 we sought to display a T cell-activating polypeptide on HIV-1 vector particles (Figure 1). Thus, an scFv derived from the OKT3 monoclonal antibody, which recognizes and activates the TCR (TCR/CD3/ζ complex, named TCR herein), was fused to the amino terminus of the SU subunit of the MLV envelope glycoprotein (Figure 1). The position of insertion of the OKT3-derived scFv was chosen to allow its functional display on virions.37,38 The resulting chimera was named OKT3SU. Because the SU and TM subunits of the MLV envelope glycoprotein are not covalently held together,41partial dissociation of the chimeric SU from the envelope complex, and thus from the viral surface, may occur.37 Therefore, a second TCR-activating chimeric glycoprotein, derived from OKT3SU, was designed to prevent the potential loss of T cell-activating polypeptides from the viral particles. In the resulting chimera, named OKT3SUx, the cleavage site between the SU and TM envelope subunits was inactivated by substitution with a noncleavable linker peptide37 (Figure 1).

The HIV-1–derived vectors were generated by transient transfections of 293T cells with plasmids encoding the viral core proteins (pCMVΔ8.2 or pCMVΔ8.91), the gene transfer vector (HPPT-EF1α-GFP), the VSV-G glycoprotein (phCMV-G), and either of the 2 TCR-activating glycoproteins (OKT3SU or OKT3SUx). Coexpression of VSV-G protein with either OKT3SU or OKT3SUx was necessary to render the viral particles fully infectious (data not shown), consistent with previous results.37 Lentiviral particles, harvested in supernatants of producer cells 2 days after transfection, were concentrated by ultracentrifugation. Incorporation of the chimeric glycoproteins on the virions was assessed by immunoblotting. Standardization of the amount of viral particles loaded on gels was determined using anticapsid antibodies (Figure2). Immunodetection of the viral pellets using antibodies against the VSV-G or MLV glycoproteins indicated that the OKT3SU and OKT3SUx chimeras were coincorporated with the VSV-G glycoprotein on the virions. The OKT3SUx chimeric glycoprotein was less abundant in the viral pellet than the OKT3SU chimera, indicating a reduced level of viral incorporation of the former glycoprotein. Differences in the electrophoretic mobilities of the OKT3SU and OKT3SUx chimeric glycoproteins demonstrated the existence of a covalent linkage between the SU and TM subunits for the OKT3SUx chimera. Compared to viral particles produced with VSV-G alone, a lower incorporation of the VSV-G glycoprotein was detected for virions generated with VSV-G and either of the 2 chimeric glycoproteins (Figure 2), most probably because of competition at the level of viral assembly.

To determine whether the OKT3-derived scFv was displayed in a correct conformation on the lentiviral particles, we performed binding and competition assays on TCR⁺ Jurkat T cells. Viral particles generated with unmodified MLV glycoproteins or with VSV-G alone were used as controls. Supernatants containing OKT3SU- and OKT3SUx-displaying viral particles were incubated with Jurkat cells at 4°C, to prevent binding to the amphotropic receptor, which is temperature-sensitive.42 Virion-to-cell binding was then analyzed by flow cytometry using antibodies against the MLV SU.38 No binding was detected with viral particles carrying the unmodified MLV glycoproteins (Figure3). In contrast, virions carrying the OKT3SU or OKT3SUx chimeras could readily bind to the target cells. Binding of virions pseudotyped with the OKT3SUx glycoprotein was lower than that of virions carrying the OKT3SU chimera, probably due to the weaker viral incorporation of the former glycoprotein (Figure 2). Competitive binding assays were performed in the presence of TCR-blocking anti-CD3 antibodies to demonstrate the specific CD3 interaction of virions displaying the OKT3-derived scFv. Almost complete inhibition of virion binding was found (Figure 3). Altogether, these results demonstrate that the OKT3-derived scFv was correctly displayed on lentiviral vector particles and could specifically allow virion binding to TCR⁺ cells.

Infectivity of lentiviral vectors was then determined on 293T target cells. Despite the presence of chimeric glycoproteins that reduced the incorporation of VSV-G (Figure 2), infectious titers of up to 5 × 10⁶ cfu/mL were readily obtained. Simultaneous vector titrations on 293T cells were performed to evaluate the MOIs used in all PBL transduction experiments (Figures 5-8). The range of MOIs applied for PBL transduction was between 4 and 60 infectious virions per target T cell.

Primary blood lymphocytes were isolated from healthy blood donors. Cell cycle analysis and detection of the CD25, CD69, and CD71 T-cell activation markers were performed before and after transduction with the lentiviral vectors. More than 99% of the PBLs were in the G₀/G₁ phase of the cell cycle before transduction (Table 1). A low proportion of PBLs showed weak staining for the CD25 low-affinity IL-2 receptor (< 3%) and for the CD69 early activation marker (< 4%; Figure4).

These resting PBLs were incubated with lentiviral vectors copseudotyped with VSV-G and OKT3SU or OKT3SUx glycoproteins. Results obtained with PBLs derived from 7 different donors are shown in Figure5. Consistent with results of others,14,29,43 nonactivated PBLs were very poorly transduced with vectors carrying VSV-G alone (< 1.5% GFP⁺ cells; mean, 0.51% ± 0.43%; n = 7). However, infections performed with the same vector particles in the presence of soluble anti-CD3 antibodies resulted in higher levels of gene transfer (Figure 5), ranging from 0.5% to 15% (mean, 7.18% ± 4.4%; n = 6), which is due to partial activation of the target cells (Figure 4). Activation with soluble anti-CD3 plus anti-CD28 antibodies rendered the PBLs highly susceptible to infection with the VSV-G–pseudotyped lentiviruses (29%-79% GFP⁺ cells; mean, 47.89% ± 19.7%; n = 7). Thus, consistent with results of others,14,43 2 stimulation signals were necessary to achieve efficient transduction of PBLs with unmodified lentiviral vectors. In contrast, in the absence of anti-CD3 and anti-CD28 antibodies, infection of PBLs with OKT3SU-displaying lentiviral vector particles resulted in very efficient transduction, ranging from 30% to 48% GFP⁺ cells (mean, 36.0% ± 6.45%; n = 7; Figure 5). This exceeded the performance of unmodified VSV-G–pseudotyped HIV-1 vectors by more than 100-fold and was even superior to that observed after addition of anti-CD3 antibodies (16-fold). Additionally, the transduction efficiency of the OKT3SU-displaying lentiviral vectors was frequently in the same range as that obtained with unmodified lentiviral vectors used in combination with anti-CD3 and anti-CD28 soluble antibodies, despite the 2- to 10-fold higher MOIs of the latter vectors (Figure 5).

The MLV SU is not covalently associated to the envelope glycoprotein complex and hence may dissociate from the surface of viral particles onto which it is incorporated.44 Therefore, efficient PBL transduction with the OKT3SU-displaying lentiviral vectors could be due either to virion-associated TCR-activating polypeptide or, alternatively, to soluble OKT3SU, “shed” from the viral particle. To discriminate between these 2 possibilities, transduction experiments of nonactivated PBLs were performed with vectors generated either with the OKT3SU glycoprotein or with the OKT3SUx chimera (Figure 1), which was engineered to avoid loss of SU by shedding. Vectors generated with either type of chimeric glycoproteins could similarly activate the resting PBLs, as judged by their capacity to up-regulate the expression of the CD25, CD69, and CD71 activation markers (Figure 4). Importantly, both vector types could efficiently transduce freshly isolated PBLs (Figure 6), though with a slightly lower efficiency for OKT3SUx-displaying vectors (37% ± 7.21% versus 24.17% ± 10.54 GFP⁺ cells), perhaps owing to a lower density of the latter chimeric glycoprotein on the viral surface (Figure 2).

Cells transduced with either OKT3SU- or OKT3SUx-displaying HIV-1 vectors were CD3⁺ as determined by counterstaining of the GFP⁺ cells with anti-CD3 antibodies (data not shown), thus indicating that the vectors could specifically and efficiently infect T lymphocytes without exogenous stimuli. Additionally, the GFP⁺ T cells exhibited proportions of CD4⁺ and CD8⁺ T cells identical to that of the control untransduced T-cell population (data not shown). Altogether these results indicate that the engineering of the viral surface of lentiviral vectors with T cell-activating polypeptides can overcome the infection block of nonactivated PBLs and allow efficient gene delivery in both CD4⁺ and CD8⁺ T cells.

The cell cycle analysis of the PBLs transduced with the different lentiviral vectors showed that no or only poor progression in the cell cycle could be detected for PBLs incubated with unmodified lentiviral vectors in the presence or in the absence of soluble anti-CD3 antibodies during infection (Table 1). In contrast transduction of PBLs in the presence of both soluble anti-CD3 and anti-CD28 antibodies resulted in enhanced proportion of cells in S/G₂M phases (up to 23% of cells). The OKT3SU- and OKT3SUx-displaying lentiviral vectors also induced cell cycle progression of infected PBLs (9.5% and 8.7% cells in S/G₂M, respectively), though at a lower extent compared to when both anti-CD3 plus anti-CD28 antibodies were added during infection (Table 1). The finding that infection with the OKT3SU-displaying vectors induced progression into the cell cycle formally raised the possibility that their high transduction efficiency could be linked to proliferation of the PBLs. To address this question, side-to-side transduction experiments of primary PBLs were performed using either lentiviral vectors or MLV-derived vectors, whose integration in target cell genome is absolutely dependent on cell proliferation.7,8 Compared to OKT3SU-diplaying lentiviral vectors, MLV vectors generated with the OKT3SU chimera incorporated similar levels of chimeric glycoproteins and had equivalent titers on 293T target cells (data not shown). Moreover, the latter vectors could induce weak cell cycle progression (Table 1) and up-regulated CD25, CD69, and CD71 markers in a manner similar to the OKT3SU-displaying lentiviral vectors (Figure 4), thus demonstrating that the surface of both types of vector particles was similarly engineered. However, in contrast to OKT3SU-displaying lentiviral vectors, infection of nonactivated PBLs with OKT3SU-displaying MLV-derived vectors only resulted in a low level of PBL transduction, of up to 4% GFP⁺ cells (mean, 1.92% ± 1.5%; n = 5), despite comparable MOIs (Figure 6). Similar weak transduction of nonactivated PBLs was obtained with the OKT3SUx-displaying MLV-based vectors (1.17% ± 0.71% GFP⁺ cells; n = 3). This low transduction efficiency most likely reflected infection of cells having entered in the cell cycle (Table 1). Indeed MLV-derived vectors with an unmodified viral surface were unable to induce cell cycle progression (Table 1) and only resulted in marginal transduction of up to 1.1% GFP⁺ PBLs (mean, 0.56% ± 0.39%; n = 5) (data not shown). Altogether these results indicate that although a small proportion of PBL transduction could be due to activation of the cell cycle, the low cell proliferation induced by the OKT3SU-displaying lentiviral vectors was not the reason for their high transduction efficiency.

Some reports have raised the possibility that nonstructural “accessory” HIV-1 proteins may positively influence gene transfer by lentiviral vectors in nonproliferating T cells.43,45Previous experiments (Figures 5 and 6) were performed using HIV-1–based vectors generated with the pCMVΔ8.91 packaging construct, which only express the Gag-Pol viral core proteins and the Tat and Rev regulatory proteins. To evaluate the importance of HIV-1 accessory proteins in PBL transduction, OKT3SU- or OKT3SUx-displaying lentiviral vectors were generated with either the pCMVΔ8.91 packaging construct or with the pCMVΔ8.2 packaging construct, which also express the HIV-1 vif, vpu, vpr, and nefaccessory genes.31 Lentiviral vectors pseudotyped only with VSV-G glycoproteins and generated with either pCMVΔ8.91 or pCMVΔ8.2 packaging constructs gave a comparable poor transduction efficiency (< 2% GFP⁺ cells) of freshly isolated PBLs (Figure 7). In contrast, lentiviral vectors generated with either OKT3SU or OKT3SUx glycoproteins could efficiently transduce the nonactivated PBLs. However, no significant differences could be detected in transduction efficiency, whether the vectors were generated with or without the HIV-1 accessory proteins (Figure 7).

Previous transduction experiments were performed by incubating the nonactivated PBLs with the viral particles for 6 days (Figures 5-7). We then sought to determine if a shorter exposure of freshly isolated resting PBLs to the OKT3SU-displaying vectors would be sufficient to activate lymphocytes and render them susceptible to infection. PBLs were therefore exposed for only 24 hours to supernatants containing HIV-1–derived or MLV-based vectors. After transduction, cells were washed and were further incubated for 5 days in media supplemented with, or without, a low concentration of rIL-2 to prevent cell death. No significant progression in cell cycle (data not shown) was found 24 hours after infection, before addition of rIL2, in contrast to previous experiments where cells had been put in contact with OKT3SU-displaying vectors for 6 days (Table 1). Almost no transduction was detected following infection with either lentiviral vectors displaying only VSV-G glycoproteins (0.67% ± 0.65%; n = 5) or with MLV-derived vectors, whether the surface of the latter vector particles carried (0.63% ± 0.53; n = 2), or not (0.25% ± 0.18%; n = 5), OKT3SU chimeric glycoproteins (Figure 8). Compared to vectors generated with unmodified glycoproteins, infection of nonactivated PBLs with OKT3SU-displaying lentiviral vectors resulted in an average 20-fold higher transduction efficiency, in the range of 9% to 19% GFP⁺ cells (mean, 12.60% ± 4.63%; n = 5). Similar results were obtained in experiments performed in the absence of rIL-2, yet the prolonged time of culture of PBLs without cytokines resulted in extensive cell death (data not shown). In conclusion, these data indicated that transduction of primary lymphocytes by a short exposure to OKT3SU-displaying lentiviral vectors gave rise to high transduction levels, which were 34-fold and 100-fold higher than those obtained with the unmodified HIV-1–derived and MLV-derived vectors, respectively.

Lentiviral vectors have shown promise in the transduction of several resting cell types such as retinal cells, pancreatic islets, cells of the central nervous system, or progenitor and differentiated hematopoietic cells.12 For these reasons, lentiviral vectors should be preferred gene delivery vehicles over vectors derived from oncoretroviruses such as MLVs that cannot transduce nonproliferating target cells.8 There are, however, important gene transfer restrictions to some nonproliferative tissues or cell types and recent studies have shown that progenitor hematopoietic stem cells in G₀, nonactivated primary blood lymphocytes or monocytes were not transducible by HIV-1–derived vectors.17,19,25 In contrast to the mechanisms that restrict gene transfer into monocytes,18,19 the 2 former cell types are refractory to gene transfer most probably because their low metabolic state prevents post-entry replication steps such as initiation or processivity of reverse-transcription and nuclear import.17,25 Thus, activation of these cells, causing G₀-to-G_1b transition of the cell cycle, is required to relieve from the blocks in gene delivery.17,25This can be achieved by adding a minimal combination of cytokines into the infection media during a short culture period.14,17,25,46 Therefore, on such minimal ex vivo infection conditions, lentiviral vectors may allow gene transfer in more primitive hematopoietic progenitor cells than those usually reached with MLV-derived vectors.46 However, there is considerable interest to further develop the lentiviral vectors so as to reach the most primitive progenitor cells that are believed to be in a quiescent state, to be slow to respond to cytokine stimulation, and to tend to lose multipotentiality or long-term repopulating capacity under cytokine stimulation. Similarly, in the case of genetic modification of T cells, most previous studies have focused on the optimization of transduction protocols, yet most gene therapy applications will require the transduction of the naive T-cell populations.14,47 It has now become evident that optimized protocols that use activation stimuli such as antibodies and mitogens may not be compatible with the preservation of the pool of naive T cells, which harbor the capacity to respond to novel antigens.4,47 

Recent improvements in the development of lentiviral vectors have focused on the optimization of nuclear import of the transgene.15,16 Inclusion of the HIV-1 central polypurine track (cPPT) in lentiviral vectors has resulted in enhanced transduction of human progenitor stem cells and T cells.14,46 However, the improved lentiviral vectors that include the cPPT sequence still fail to transduce nonactivated T lymphocytes,14 most likely because the primary block in initiation or completion of reverse transcription could not be alleviated with the novel vectors. Therefore, because minimal activation of resting human progenitor stem cells and T cells with stimuli that cause G₀-to-G₁ transition allows gene transfer,17,25,29 we sought to generate surface-modified lentiviral vector particles that would convey a transient activation signal in target cells at the time of gene delivery. As a proof of concept, we show here that lentiviral vectors that display on their viral surface an anti-CD3 scFv T cell-activating polypeptide mediate efficient gene transfer into freshly isolated lymphocytes not activated by exogenous stimuli. No addition of T cell-activating stimuli in the infection media was required to achieve transduction levels of up to 48% GFP⁺ cells. In comparison, unmodified lentiviral vectors whose viral surface expressed only the VSV-G glycoprotein were unable to transduce freshly isolated T cells.

Optimal activation of resting T cells requires signaling via the TCR and additional stimulation provided through accessory molecules present on the surface of antigen-presenting cells (APCs). For highly purified resting T cells, multivalent APC stimulation can be mimicked ex vivo with immobilized anti-CD3 antibodies and causes a G₀-to-G_1a transition. However, this does not allow completion of HIV-1 reverse transcription25 or efficient transduction with HIV-1–based vectors.29Costimulation with immobilized anti-CD3 and anti-CD28 antibodies allows progression to G_1b48,49 and allows successful infection by wild-type HIV-125 or transduction by lentiviral vectors.14 Prolonged cell exposure to high loads of immobilized anti-CD3 and anti-CD28 antibodies induces cell proliferation50 and allows transduction by vectors derived from either lentiviruses or oncoretroviruses.4,14 Our results confirm the notion that induction of cell division is not a prerequisite for efficient gene transfer in PBLs with lentiviral vectors. No HIV-1 accessory proteins, such as vpr or vif, were required to achieve high-level transduction in the PBLs, in agreement with results of others.13,14 Interestingly, our data show that comparable transduction levels can be achieved with either the surface-modified, OKT3SU-displaying, lentiviral vectors or with the unmodified vectors used in combination with both soluble anti-CD3 and anti-CD28 antibodies. This indicates that an exogenous costimulation signal may not be needed for transduction with the OKT3SU-displaying lentiviral vectors. These results, which are in contrast to the poor transduction levels obtained with VSV-G–pseudotyped lentiviral vectors in the presence of anti-CD3 antibodies alone, are likely to be explained by the fact that the multivalent presentation on viral particles of TCR-binding polypeptides may strongly favor TCR cross-linking. Additionally, local concentration of TCR-binding polypeptides could be much higher when presented on virions compared to when provided to cells as soluble or immobilized anti-CD3 antibodies. In agreement with this hypothesis, concentrations of immobilized anti-CD3 antibodies 10-fold higher than those classically used in T-cell activation and transduction experiments, that is, around 1 μg/well,14,25 were found sufficient to allow infection by HIV-1 and could also induce cell proliferation.50Alternatively, concomitant to partial activation induced by the OKT3SU-displaying virions, costimulation could be provided either by residual APCs likely to be present in the preparations of PBLs25,49 or by costimulatory molecules that may have been coincorporated on the surface of the lentiviral particles.51 

T-cell receptor cross-linking induced by surface-modified lentiviral vectors may have adverse effects on the phenotype of the transduced T cells, such as inducing anergy or modifying their ability to respond to novel antigens.52 An important component of the homeostasis of the immune system is the maintenance of the peripheral naive T-cell repertoire. Human T cells can be divided into naive and memory subsets based, respectively, on expression of RA and RO isoforms of the CD45 molecule. In contrast to memory T cells, naive cells exhibit few effectors functions (eg, cytokine production), are less susceptible to activation-induced cell death, and display robust proliferation in response to TCR-mediated activation signals. It is important that the compartment of naive CD45RA⁺ cells is maintained through life. Although TCR activation induced on infection with OKT3SU-displaying viral particles is short, the change in T-cell phenotype cannot be formally excluded. In fact, transduction of PBLs with OKT3SU-displaying lentiviral vectors was found to shift the T cells to the memory phenotype (data not shown), in agreement with results of others using TCR stimulation by anti-CD3/anti-CD28 antibodies to allow infection of resting PBLs by retroviral vectors.4 It should be pointed out that the approach reported here was designed to provide proof of concept of gene transfer strategies that may have utility elsewhere. Here, we show for the first time that it is possible to efficiently transduce quiescent PBLs by a lentiviral vector that displays a T cell-activating polypeptide. For the reasons discussed above, it might be preferable to display cytokines like IL-2, IL-7, and IL-4, or combinations of these polypeptides, on the surface of the lentiviral vectors. These cytokines are known to promote long-term survival of resting T cells while maintaining the phenotype of the subset of naive cells.53,54 Previously, we described the generation of MLV-derived vectors that displayed IL-2 chimeric envelope glycoproteins (IL-2SU).37 These vectors allowed efficient transduction of G₀/G₁-arrested cell lines expressing the IL-2 receptor.37 However, lentiviral vectors pseudotyped with VSV-G and chimeric IL-2SU envelope glycoproteins were not able to transduce freshly isolated PBLs, despite their capacity to weakly up-regulate the CD25 and CD69 activation markers (data not shown). This result was most likely due to the low expression of high-affinity IL-2 receptors on resting PBLs. Alternatively, IL-7 is a promising candidate because it is known that it activates PBLs and permits infection with lentiviral vectors into both memory and naive T cells.29,47 Moreover, IL-7 contributes to the maintenance of the adult CD45RA⁺ T-cell pool.55,56Characterization of lentiviral vectors that display some of these polypeptides on the viral surface are currently underway in our laboratory and will be valuable tools to selectively activate specific T-cell subsets, allowing numerous studies of the physiology of lymphocytes.

We are grateful to Naomi Taylor and Jacqueline Marvel for stimulating discussions and for critical reading of the manuscript.